1       SUBROUTINE DLAQSB( UPLO, N, KD, AB, LDAB, S, SCOND, AMAX, EQUED )
  2 *
  3 *  -- LAPACK auxiliary routine (version 3.3.1) --
  4 *  -- LAPACK is a software package provided by Univ. of Tennessee,    --
  5 *  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
  6 *  -- April 2011                                                      --
  7 *
  8 *     .. Scalar Arguments ..
  9       CHARACTER          EQUED, UPLO
 10       INTEGER            KD, LDAB, N
 11       DOUBLE PRECISION   AMAX, SCOND
 12 *     ..
 13 *     .. Array Arguments ..
 14       DOUBLE PRECISION   AB( LDAB, * ), S( * )
 15 *     ..
 16 *
 17 *  Purpose
 18 *  =======
 19 *
 20 *  DLAQSB equilibrates a symmetric band matrix A using the scaling
 21 *  factors in the vector S.
 22 *
 23 *  Arguments
 24 *  =========
 25 *
 26 *  UPLO    (input) CHARACTER*1
 27 *          Specifies whether the upper or lower triangular part of the
 28 *          symmetric matrix A is stored.
 29 *          = 'U':  Upper triangular
 30 *          = 'L':  Lower triangular
 31 *
 32 *  N       (input) INTEGER
 33 *          The order of the matrix A.  N >= 0.
 34 *
 35 *  KD      (input) INTEGER
 36 *          The number of super-diagonals of the matrix A if UPLO = 'U',
 37 *          or the number of sub-diagonals if UPLO = 'L'.  KD >= 0.
 38 *
 39 *  AB      (input/output) DOUBLE PRECISION array, dimension (LDAB,N)
 40 *          On entry, the upper or lower triangle of the symmetric band
 41 *          matrix A, stored in the first KD+1 rows of the array.  The
 42 *          j-th column of A is stored in the j-th column of the array AB
 43 *          as follows:
 44 *          if UPLO = 'U', AB(kd+1+i-j,j) = A(i,j) for max(1,j-kd)<=i<=j;
 45 *          if UPLO = 'L', AB(1+i-j,j)    = A(i,j) for j<=i<=min(n,j+kd).
 46 *
 47 *          On exit, if INFO = 0, the triangular factor U or L from the
 48 *          Cholesky factorization A = U**T*U or A = L*L**T of the band
 49 *          matrix A, in the same storage format as A.
 50 *
 51 *  LDAB    (input) INTEGER
 52 *          The leading dimension of the array AB.  LDAB >= KD+1.
 53 *
 54 *  S       (input) DOUBLE PRECISION array, dimension (N)
 55 *          The scale factors for A.
 56 *
 57 *  SCOND   (input) DOUBLE PRECISION
 58 *          Ratio of the smallest S(i) to the largest S(i).
 59 *
 60 *  AMAX    (input) DOUBLE PRECISION
 61 *          Absolute value of largest matrix entry.
 62 *
 63 *  EQUED   (output) CHARACTER*1
 64 *          Specifies whether or not equilibration was done.
 65 *          = 'N':  No equilibration.
 66 *          = 'Y':  Equilibration was done, i.e., A has been replaced by
 67 *                  diag(S) * A * diag(S).
 68 *
 69 *  Internal Parameters
 70 *  ===================
 71 *
 72 *  THRESH is a threshold value used to decide if scaling should be done
 73 *  based on the ratio of the scaling factors.  If SCOND < THRESH,
 74 *  scaling is done.
 75 *
 76 *  LARGE and SMALL are threshold values used to decide if scaling should
 77 *  be done based on the absolute size of the largest matrix element.
 78 *  If AMAX > LARGE or AMAX < SMALL, scaling is done.
 79 *
 80 *  =====================================================================
 81 *
 82 *     .. Parameters ..
 83       DOUBLE PRECISION   ONE, THRESH
 84       PARAMETER          ( ONE = 1.0D+0, THRESH = 0.1D+0 )
 85 *     ..
 86 *     .. Local Scalars ..
 87       INTEGER            I, J
 88       DOUBLE PRECISION   CJ, LARGE, SMALL
 89 *     ..
 90 *     .. External Functions ..
 91       LOGICAL            LSAME
 92       DOUBLE PRECISION   DLAMCH
 93       EXTERNAL           LSAME, DLAMCH
 94 *     ..
 95 *     .. Intrinsic Functions ..
 96       INTRINSIC          MAXMIN
 97 *     ..
 98 *     .. Executable Statements ..
 99 *
100 *     Quick return if possible
101 *
102       IF( N.LE.0 ) THEN
103          EQUED = 'N'
104          RETURN
105       END IF
106 *
107 *     Initialize LARGE and SMALL.
108 *
109       SMALL = DLAMCH( 'Safe minimum' ) / DLAMCH( 'Precision' )
110       LARGE = ONE / SMALL
111 *
112       IF( SCOND.GE.THRESH .AND. AMAX.GE.SMALL .AND. AMAX.LE.LARGE ) THEN
113 *
114 *        No equilibration
115 *
116          EQUED = 'N'
117       ELSE
118 *
119 *        Replace A by diag(S) * A * diag(S).
120 *
121          IF( LSAME( UPLO, 'U' ) ) THEN
122 *
123 *           Upper triangle of A is stored in band format.
124 *
125             DO 20 J = 1, N
126                CJ = S( J )
127                DO 10 I = MAX1, J-KD ), J
128                   AB( KD+1+I-J, J ) = CJ*S( I )*AB( KD+1+I-J, J )
129    10          CONTINUE
130    20       CONTINUE
131          ELSE
132 *
133 *           Lower triangle of A is stored.
134 *
135             DO 40 J = 1, N
136                CJ = S( J )
137                DO 30 I = J, MIN( N, J+KD )
138                   AB( 1+I-J, J ) = CJ*S( I )*AB( 1+I-J, J )
139    30          CONTINUE
140    40       CONTINUE
141          END IF
142          EQUED = 'Y'
143       END IF
144 *
145       RETURN
146 *
147 *     End of DLAQSB
148 *
149       END